Machine learning based scatter correction

Abstract

The present approach relates to the use of machine-learning in convolution kernel design for scatter correction. In one aspect, a neural network is trained to replace or improve the convolution kernel used for scatter correction. The training data set may be generated probabilistically so that actual measurements are not employed.

Description

TECHNICAL FIELD[0001]The subject matter disclosed herein relates to scatter correction in imaging contexts, and in particular to the use of machine learning techniques to facilitate scatter artifact correction or reduction.BACKGROUND[0002]Non-invasive imaging technologies allow images of the internal structures or features of a patient / object to be obtained without performing an invasive procedure on the patient / object. In particular, such non-invasive imaging technologies rely on various physical principles (such as the differential transmission of X-rays through a target volume, the reflection of acoustic waves within the volume, the paramagnetic properties of different tissues and materials within the volume, the breakdown of targeted radionuclides within the body, and so forth) to acquire data and to construct images or otherwise represent the observed internal features of the patient / object.[0003]Certain such imaging techniques, including Positrons Emission Tomography (PET), X-...

AI Technical Summary

Benefits of technology

The patent text discusses a method called convolution kernel based deconvolution for computed tomography (CT) and X-ray systems. This method estimates the scatter profile by convolving with spatially invariant kernel models, but its accuracy depends on the kernel design and parameters. The kernel parameters can vary from scan settings to another, so it is challenging to design an accurate kernel and determine its parameters. The technical effect of the patent text is to provide a solution for accurately estimating the scatter profile in X-ray based systems using the deconvolution method.

Problems solved by technology

This method does not require additional hardware or scans, and the computational cost is typically low, especially compared to high-cost Monte-Carlo-based scatter simulation.

Therefore, it is a challenge to design a kernel and determine the kernel parameters to accurately estimate the scatter profile.

Images